During Google I/O 2025, Google announced that it is deploying Project Mariner, the company’s experimental AI agent that browses and uses websites, to more users and developers. Google also has significantly updated how Project Mariner works, saying that agents can take on almost a dozen tasks at a time.
US subscribers of Google’s new $249.99/month AI Ultra plan have access to Project Mariner, and the company says support for more countries is coming soon. Google also says it brings Project Mariner’s capabilities to Gemini API and Vertex AI, allowing developers to build agents-equipped applications.
First unveiled in late 2024, Project Mariner represents Google’s boldest effort to revamp the way users interact with the Internet via AI agents. At launch, Google Search leaders said they viewed Project Mariner as part of a fundamental user experience shift. In this shift, instead of visiting a website and completing tasks, you delegate more tasks to an AI agent.
For example, Project Mariner users can purchase tickets for baseball games and buy groceries online without having to access third-party websites. Simply chat with Google’s AI agent to visit our website and take actions.
Project Mariner competes with other web browsing AI agents, including Openai operators, Amazon’s Nova Act, and Anthropic’s computer use. All of these tools are in the experimental stage, and TechCrunch’s experience has proven to be slow in prototypes and tend to be misleading.
However, Google says it received feedback from early testers to improve Project Mariner’s capabilities. A Google spokesman tells TechCrunch to update Project Mariner to run on virtual machines in the cloud, just like Openai and Amazon agents. This means that users can work on other projects, but Project Mariner completes tasks in the background. Google says the new Project Mariner can handle up to 10 tasks at the same time.
This update makes Project Mariner more useful compared to its predecessors run in the user’s browser. As mentioned in the first review, the early design of Project Mariner meant that users could not use other tabs or apps on the desktop while the AI agent was working. This kind of purpose has defeated the purpose of AI agents.
Over the next few months, Google says users will be able to access Project Mariner in AI mode, a Google search experience powered by AI. Upon launch, this feature is limited to Search Lab, Google’s opt-in test ground for search functionality. Google says it is working with Ticketmaster, StubHub, Resy and Vagaro to enhance some of these agent flows.
Separately, Google has announced an early demo of another agent experience called “agent mode.” According to the company, the feature combines web browsing with research features, integrations and other Google apps. Google says Ultra subscribers can access Agent mode immediately on their desktop.
With this year’s I/O, Google seems willing to finally be willing to ship the agent experience it has been talking about for years. Project Mariner, Agent Mode, and AI Mode all appear to be poised to change the way users navigate the web and how vendors interact with customers online. Web browsing agents have a big impact on the internet economy, but Google seems ready to put all these agents in the world.